Online support vector machine based on linear independent

In this paper a novel online learning algorithm for support vector machine is proposed. In this algorithm support vector are extracted based on the property of linear independent. Error estimation can be quickly got by using least square support vector machine. Then learning scheme is executed or not depended on the classification error. In the learning scheme support vectors are re-selected and re-training according to linear independent and WLS-SVM. Experimental results reveal that our algorithm outperforms IncrSVM and LIBSVM on the time complexity and classification rate.